In this chapter, we use the local surface patch (LSP) representation for matching 3D ears [1, 12]. The LSP representation, a new local surface descriptor, is characterized by a centroid, a local surface type and a 2D histogram. The 2D histogram shows the frequency of occurrence of shape index values vs. the angles between the normal of reference feature point and that of its neighbors. The proposed human recognition system using the LSP representation is illustrated in Figure 5.1. The local surface descriptors are computed for the feature points which are defined as either the local minimum or the local maximum of shape indexes. By comparing the local surface patches for a gallery and a probe image, the potential corresponding local surface patches are established and then filtered by geometric constraints. Based on the filtered correspondences, the initial rigid transformation is estimated. Once this transformation is obtained, it is then applied to randomly selected control points of the hypothesized gallery ear in the database. A modified iterative closest point (ICP) algorithm is run to improve the transformation which brings a gallery ear and a probe ear into the best alignment, for every gallery-probe pair. The root mean square (RMS) registration error is used as the matching error criterion. The subject in the gallery with the minimum RMS error is declared as the recognized person in the probe image.
Access provided by Autonomous University of Puebla. Download to read the full chapter text
Chapter PDF
Rights and permissions
Copyright information
© 2008 Springer-Verlag London Limited
About this chapter
Cite this chapter
(2008). Recognizing 3D Ears Using Local Surface Patches. In: Human Ear Recognition by Computer. Advances in Pattern Recognition. Springer, London. https://doi.org/10.1007/978-1-84800-129-9_5
Download citation
DOI: https://doi.org/10.1007/978-1-84800-129-9_5
Publisher Name: Springer, London
Print ISBN: 978-1-84800-128-2
Online ISBN: 978-1-84800-129-9
eBook Packages: Computer ScienceComputer Science (R0)